900,000 reasons for working in AI 🤑

Backed by the Chief Sumo, Noah Kagan, Castmagic helps you turn productive meetings into “How-to” guides for your team, and podcasts into twitter threads. Turn your audio into content like magic with Castmagic.

AI still needs humans. Welcome to the AllThingsAI newsletter. Let’s get into it.

Breakdown:

• Companies are shelling out heaps of cash for AI skills.

• Red-teaming AI.

• AI can listen in on your typing.

• The New York Times wants you to stop scraping its pages.

• Big Blue is working on a more efficient AI chip.

• And we’ve got an entire stack of great AI tools for you to check out.

Wanted: AI talent

While much of the commentary around AI has been about how such technologies might decimate jobs, there will, it seems, still be winners.

As The Wall Street Journal (paywall) reports, the crush for AI expertise is pushing salaries to eye-watering levels.

Dating app Hinge is looking for a vice president of AI who could earn almost $400,000 a year, while the freelancer platform Upwork is offering close to $440,000 annually for a similar role.

But for those AI practitioners looking to purchase an even bigger yacht, Netflix is offering as much as $900,000 a year for a machine learning product manager.

NINE-HUNDRED-THOUSAND-DOLLARS. A YEAR.

You know what $900,000 can buy you? ChatGPT told me it’s enough for 20,000 pairs of Crocs. Perfect to go with that new boat.

Of course, with such sky-high compensations sloshing around, that naturally puts pressure on other companies who simply cannot pony up almost a million dollars a year for an AI expert.

The report quotes Chris Todd, CEO at software business UKG, who notes the demand for AI talent is “smoking hot.”

Meanwhile, Paul J. Groce, the head of recruitment firm Leathwaite makes a succinct point about the state of play: “We do not magically have thousands of additional AI developers, product managers, and everything else.”

So, if you’re worried about AI flattening your industry, perhaps now’s the time to learn how to drive a metaphorical AI steamroller. And then a (non-metaphorical) boat, of course.

Why it matters:
Companies don’t pay more for talent than they have to - especially those that are beholden to shareholders. So Netflix’s decision to offer $900,000 for a single AI head is perhaps a sign of how much FOMO all these big operators are feeling. But with AI salaries nearing seven digits, that will surely encourage more people to focus on the field, meaning compensations likely won’t stay so high forever.

Tome isn’t a deck, or a doc, or a splash page. It’s a web-based multimedia canvas built for how we communicate today.

Tome’s AI tools help you generate new content so you can get your point across compellingly. Get started for free today.

DEF CON gets into AI red-teaming

Thousands of hackers showed up in Las Vegas this past weekend, determined to bend machines to their will. But rather than being some elaborate slot machine heist, the hackers were in town for DEF CON.

For the uninitiated, DEF CON is an annual conference that invites hackers from around the globe and tasks them with finding novel vulnerabilities across a whole array of systems and services. This includes products from the likes of Google, Microsoft, Nvidia, and more. Challenges are arranged in “villages,” with each attracting its own group of specialists.

With AI being the buzzword of the moment, it’s perhaps no surprise that this year’s DEF CON featured an AI village. Here, hackers were given specific tasks designed to test the fences of various AI tools - a process known colloquially as “red-teaming.” (Personally, I think “big scary hack attack” is a cooler phrase, but I don’t make the rules.)

The first day of DEF CON saw long lines as hackers eagerly waited their turn to bust some AI systems and make them do naughty things. But, as Axios reports, the results of those efforts were somewhat mixed.

One set of hackers were able to use generative AI to build a nefarious ecommerce site in about the time it takes to boil an egg. But another prompt engineer who got an AI platform to explain how to stalk a person found the system refused such a request the day just a day later. (I guess we just have to hope bad guys are terrible note-takers or have short memory spans.)

Why it matters:
While this was the first AI village at DEF CON, the pace and scale of AI innovation means it’s likely to become a regular feature at the event moving forward. And considering there are AI doomsayers both outside and inside the industry (see: OpenAI’s Sam Alman’s statement about AI possibly going “quite wrong”), the importance of white-hat hackers finding flaws cannot be overstated.

Results from our last poll | Could deep fake artists be the end of musical creativity?

🟩🟩🟩🟩🟩🟩 Yes - there’s no replacing the real thing 🎤

⬜️⬜️⬜️⬜️⬜️⬜️ No - it’s a great way of building on a musician’s artistry 🤘

🟨⬜️⬜️⬜️⬜️⬜️ Please. Autotune ruined music long before AI showed up 🎹 

In the wild | Stories worth reading

ArsTechnica | AI researchers claim 93% accuracy in detecting keystrokes over Zoom audio

Bad guys have long used key-logging malware to hack individuals and organizations. But now researchers in the UK have demonstrated how an AI model can interpret audible keyboard clicks via teleconferencing calls to work out what is being typed. Perhaps something to be conscious of the next time you’re stuck on a boring Zoom call and find yourself tapping away on Reddit.

Adweek | ​​The New York Times Updates Terms of Service to Prevent AI Scraping Its Content

The Gray Lady has changed its terms of service (ToS), specifying that its content cannot be scraped by third-parties for large language models and other AI tools. The New York Times says unauthorized data gathering - which includes images, videos, and audio - could result in penalties. There’s no clarification of what those penalties are, but as anyone who has spent time in the city knows, you always think twice before messing with a New Yorker.

BBC | Prototype 'Brain-like' chip promises greener AI, says tech giant

IBM says it has developed a prototype chip that requires fewer resources when processing AI data. The company behind Watson (ostensibly the OG of AI) explains the chip makes use of memristors, making it analogous to how the “biological brain” creates internal networks. IBM suggests the prototype could lead the way to more energy-efficient chips for smartphones and other battery-reliant devices.

AllThingsAI | Best Bets

Did you enjoy today’s AllThingsAI newsletter? Why not forward it to someone who would also appreciate it.

Someone forwarded you this AllThingsAI newsletter? Sign up for this FREE newsletter right here so you never miss an edition.

Advertise with AllThingsAI to get your startup in front of VCs, investors and 10,000 of organic readers - not paid. Get in touch today…